A smart method for fault diagnosis modeling of automotive components, electronic equipment, and software products.

By constructing a fault diagnosis method for automotive components based on a time series model, and utilizing historical vehicle data and artificial intelligence algorithms, the method enables the prediction and early warning of faults that cannot be detected by sensors. This solves the problems of time-consuming and low-accuracy diagnosis in existing technologies, and improves the accuracy and efficiency of fault diagnosis.

CN115660640BActive Publication Date: 2026-11-17CHONGQING CHANGAN AUTOMOBILE CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202211296026.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-21
Publication Date
2026-11-17
Estimated Expiration
2042-10-21

AI Technical Summary

Technical Problem

Existing technologies are time-consuming and labor-intensive in diagnosing automotive component faults, have low diagnostic accuracy, cannot provide early warnings, and are difficult to detect faults that sensors cannot detect.

Method used

Based on vehicle model data, component data, maintenance records, and driving behavior data, a time series model is constructed using artificial intelligence algorithms to perform data cleaning, aggregation, and labeling of automotive component failures, thereby enabling failure prediction and early warning.

Benefits of technology

It improves the accuracy and efficiency of fault diagnosis, avoids misdiagnosis and incorrect replacement, reduces user complaints, and enhances the repair timeliness and customer satisfaction of 4S stores.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115660640B_ABST
    Figure CN115660640B_ABST
Patent Text Reader

Abstract

This invention provides an intelligent method and electronic device program for fault diagnosis and modeling of automotive components. The method includes acquiring historical big data on vehicle driving history and maintenance records; aggregating and cleaning historical data throughout the entire lifecycle of automotive components; segmenting and labeling data using a single start-stop cycle during vehicle operation as a time sliding window; and constructing a fault diagnosis model based on an LSTM time-series model and a neural network model. This invention uses a data-driven approach and artificial intelligence algorithms to detect and predict faults in automotive components during driving. It can diagnose and predict faults in components that cannot be detected by sensors during vehicle operation, enabling early warning of faults and improving the accuracy and timeliness of fault repair at 4S dealerships, thus enhancing fault diagnosis efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of automotive fault diagnosis and repair, specifically involving methods for fault diagnosis and modeling of automotive parts. Background Technology

[0002] As the automotive industry enters an era of intelligentization, electrification, and autonomous driving, the integration and automation of vehicle components are increasing, leading to greater demands for monitoring the operational status and health of these components. Traditional automotive component fault diagnosis relies primarily on professional repair personnel using specialized equipment to systematically troubleshoot and test fault sources. While repair personnel can resolve most minor issues, they inevitably encounter complex problems requiring expert guidance. In such cases, on-site diagnosis by OEM technical experts is necessary, which is time-consuming and labor-intensive. Furthermore, fault detection is only available offline after a fault occurs, lacking early warning capabilities. Currently, in the context of automotive intelligence, utilizing artificial intelligence technology for automotive component fault diagnosis has become possible.

[0003] Patent document 202111446107.5 discloses a configurable fault diagnosis method and system based on real-time vehicle conditions. According to the vehicle series information and real-time vehicle condition data of the vehicle under test, the basic fault detection items corresponding to the vehicle series information and the vehicle condition configuration fault detection items corresponding to the real-time vehicle condition data are dynamically configured. Then, the fault diagnosis detection is performed according to the fault diagnosis detection rules corresponding to each of the configured fault detection items. This patent relies on a predefined fault diagnosis rule database. Its fault diagnosis detection types and methods are fixed and cannot adapt to changes in automotive component faults.

[0004] Patent document 201310541642.8 discloses a vehicle condition real-time diagnostic system and its diagnostic method, which uses vehicle OBD fault code data to perform vehicle fault diagnosis. However, it is limited to the existing detection types and methods of vehicle OBD fault diagnosis and detection, and cannot detect components that are not included in the OBD fault diagnosis and detection types.

[0005] It is evident that current technologies cannot adequately address the problems in automotive repair, such as the time-consuming and labor-intensive process of detecting and diagnosing component faults, low diagnostic accuracy, and the inability of sensors to detect faults while the vehicle is in motion. Summary of the Invention

[0006] This invention addresses the shortcomings of existing technologies by proposing an intelligent method for diagnosing and modeling automotive component faults. Based on data such as vehicle model data, component data, maintenance records, historical big data from vehicle operation, and user driving behavior data, this data-driven approach utilizes artificial intelligence algorithms to detect and predict faults in automotive components during operation. It can diagnose and predict faults in components that cannot be detected by sensors during vehicle operation, enabling early warning of faults and improving the accuracy and timeliness of 4S shop fault repairs. This enhances fault diagnosis efficiency, avoids misdiagnosis, incorrect replacement, and incorrect component replacement, and reduces user complaints about faults.

[0007] Based on the above methods, the present invention also provides an electronic device and computer program product for intelligent automotive component fault diagnosis modeling.

[0008] To solve the technical problems in this invention, the following technical solution is adopted: The present invention proposes an intelligent method for fault diagnosis and modeling of automotive components in its first aspect, which mainly includes the following steps: Step 1: Obtain vehicle driving history big data and maintenance record data Based on the maintenance record cards of each vehicle component, the fault type and corresponding repair time of each component are determined. Based on the fault type and repair time, historical data of the entire life cycle of the components during the vehicle's operation from the vehicle's manufacturing time to the repair time are extracted from the cloud platform. The data fields include CAN signal data such as the component itself, component load, vehicle driving external environment, and vehicle driving behavior.

[0009] Step 2: Aggregate and clean historical data throughout the entire lifecycle of the components. The CAN signal data of each data field of the component is transmitted through different CAN buses. The CAN signal parameters of different data fields have different sampling intervals and sampling periods, and the acquisition frequency of each signal is relatively high, with the signal data being millisecond-level data. This invention is based on time series model for fault diagnosis modeling, which requires data cleaning and data aggregation processing of the extracted historical data of the entire life cycle of the component.

[0010] The first step is data cleaning, which includes removing outliers and filling in missing values ​​for signals in various data fields of historical data.

[0011] Next is data aggregation. To prevent a large amount of data loss and excessively long time series, data aggregation is performed in minutes. The signals of each data field in the historical data are sorted in chronological order and grouped in minutes. The average value of each minute group is taken as the value of the current minute.

[0012] Step 3: Time Sequence Segmentation and Labeling One start-stop cycle (from starting the car to turning it off) during vehicle operation constitutes one loop. This invention uses one start-stop cycle as a time sliding window for data segmentation and labeling, as follows: First, the data aggregated in step 2 is sorted in chronological order, and then grouped by single start and stop times.

[0013] Secondly, since different drivers have different driving habits during vehicle operation, the start-stop cycle is different each time, that is, the time series length of each single start-stop is different. It is necessary to adjust the time series length of each single start-stop to be consistent. Based on the driving habits of most users, this invention fixes the time series length of a single start-stop to a fixed value L. When the time series length of a single start-stop exceeds L, only the data from the start time to time L is extracted as the time series data of this start-stop. When the time series length of a single start-stop does not exceed L, the time series data from the start time to the stop time remains unchanged, while the time series from the stop time to time L is filled with 0 values.

[0014] Finally, the data from each start-stop operation is labeled accordingly.

[0015] Step 4: Component Fault Diagnosis Modeling Based on Time Series Model For non-accidental and non-sudden fault types, there is a slow change process in the time dimension. This invention uses a time series model for modeling and analysis, and sends the time segment data generated in step 3 into the constructed time series model for training and iteration.

[0016] The beneficial effects of the present invention using the above technical solution are as follows: 1. This invention utilizes vehicle model data, component data, maintenance record data, and big data on vehicle driving history to construct a fault diagnosis model for automotive components. This data source is comprehensive and rich, including a large amount of data that cannot be detected by sensors, enabling more accurate and comprehensive prediction of automotive component faults. In particular, this model can predict faults that cannot be detected by sensors and predict faults before they occur, achieving early warning, rapid fault location, improving the timeliness and accuracy of fault repair in 4S stores, avoiding misdiagnosis, incorrect replacement, and incorrect replacement of components, reducing user complaints about faults, and improving customer satisfaction.

[0017] 2. The initial time of component failure should be some time before the repair time. However, in reality, it is difficult to determine the initial time of component failure, which brings certain difficulties to data labeling. This invention aggregates data in minutes and performs data sliding windowing and labeling on the aggregated data. This method can solve the problem of data label accuracy during data labeling and ensure the accuracy and precision of the model in the modeling process.

[0018] In a second aspect, the present invention also provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform the intelligent automotive component fault diagnosis modeling method according to the first aspect.

[0019] Furthermore, in a third aspect, the present invention also provides a computer program product, comprising a computer program, wherein, when executed by a computer's processor, the computer program is used to cause the computer to perform the intelligent automotive component fault diagnosis modeling method according to the first aspect.

[0020] The above-mentioned electronic devices and computer program products have the same technical effects as the first aspect of this invention. Attached Figure Description Figure 1 This is a flowchart of the fault diagnosis process of the present invention; Figure 2 This is a simplified diagram of the data sliding window and annotation during data processing in this invention; Figure 3 This is a flowchart of the time series length alignment process of the present invention; Figure 4 The flowchart for modeling the time series model of this invention is shown. Detailed Implementation Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0021] It should be understood that the steps described in the method embodiments of this application may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this application is not limited in this respect.

[0022] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this application are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0023] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] The names of the messages or information exchanged between multiple units in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0025] See Figure 1 In one embodiment, a fault diagnosis modeling process for automotive components is illustrated. Taking a battery fault diagnosis model as an example, the battery driving history data includes signal data such as battery current, voltage, SOC, SOH, battery temperature, vehicle speed, ambient temperature, headlight signals, and air conditioning status. Since the original data is sampled at the millisecond level, the original data is aggregated by minute, retaining the average value of each signal within each minute. Finally, the data is sorted in chronological order, with each start-stop cycle (from vehicle startup to vehicle shutdown) as a time series. The data is then input into the fault diagnosis model, and a set of diagnostic results is output.

[0026] The process is explained in detail below: Step 1: Obtain vehicle driving history big data and maintenance record data.

[0027] Based on the maintenance record cards of each vehicle component, the fault type and corresponding repair time of each component are determined. Based on the fault type and repair time, historical data of the entire life cycle of the components during the vehicle's operation from the vehicle's manufacturing time to the repair time are extracted from the cloud platform. The data fields include CAN signal data such as the component itself, component load, vehicle driving external environment, and vehicle driving behavior.

[0028] Taking battery fault diagnosis as an example, battery driving history data includes signal data such as battery current, voltage, SOC, SOH, battery temperature, vehicle speed, external ambient temperature, vehicle light signals, and air conditioning status.

[0029] Step 2: Aggregate and clean historical data of the entire life cycle of automotive parts.

[0030] The first step is data cleaning, which includes removing outliers and filling in missing values ​​for signals in various data fields of historical data.

[0031] Next is data aggregation. To prevent a large amount of data loss and excessively long time series, since the original data is sampled at the millisecond level, the original data is aggregated at the minute level. The signals of each data field of the historical data are sorted in chronological order and grouped at the minute level. The average value of each minute group is taken as the value of the current minute.

[0032] Step 3: Segment and label the data using a single start-stop cycle during the car's operation as a time window.

[0033] First, the data aggregated in step 2 is sorted in chronological order, and then the data is segmented using a single start-stop cycle during the car's driving process as a time sliding window.

[0034] In one embodiment, see Figure 2 This invention uses a sliding window and annotation diagram for data processing. For non-accidental, non-sudden automotive component failures, the changes occur slowly over time. However, when a car malfunctions and is repaired at a 4S dealership, the repair time recorded on the maintenance record card is not the initial time when the component's failure symptoms appeared. The initial time should be some time before the repair time. However, in reality, it is difficult to determine the initial time of the component's failure symptoms, which poses a challenge to data labeling. To address the accuracy of data labels during data labeling, this invention sets three time windows (T1, T2, and T3) within the component's full lifecycle historical data (from purchase time to repair time). Data in time window T1 is labeled with a Label, while data in time window T3 is labeled with the corresponding Label' based on the specific failure type. Data in time window T2, which contains both Label and Label' data, is discarded. This ensures both the accuracy of data labels and a sufficient sample size for failure data, avoiding problems such as imbalanced positive and negative samples.

[0035] Within time windows T1 and T3, the data is processed and labeled using a single vehicle start-stop cycle (from vehicle start to vehicle shutdown) as the sliding time window.

[0036] Next, adjust the time sequence length of each single start-stop time series data to a uniform fixed length L: fix the time sequence length of a single start-stop to a fixed value L, determine whether the length N of each single start-stop time series data is greater than the set fixed value L. When N>L, only the data from the start time to the L-th time is intercepted as the single start-stop time series data of this time; when N<L, the time series data from the start time to the stop time remains unchanged, and the time series from the stop time to the L-th time is filled with 0 values. For the time sequence length alignment process, refer to Figure 3 .

[0037] Finally, assign corresponding labels to the single start-stop data.

[0038] Step 4: Construct a fault diagnosis model based on the LSTM time sequence model and the neural network model, and finally input the processed data into the built algorithm model by single start-stop time series for training and verification, refer to Figure 4 .

[0039] The above modeling method for automotive component fault diagnosis can be used for fault diagnosis and prediction of faults that cannot be perceived by sensors during vehicle driving, so as to realize fault diagnosis and early warning of each component and prevent accidents.

[0040] Further, the exemplary embodiments of the present application further provide an electronic device, comprising: at least one processor; and a memory communicatively connected with the at least one processor. The memory stores a computer program executable by the at least one processor, and the computer program, when executed by the at least one processor, is configured to cause the electronic device to execute the modeling method for automotive component fault diagnosis presented in the above embodiments of the present application.

[0041] The exemplary embodiments of the present application further provide a computer program product, comprising a computer program, wherein the computer program, when executed by a processor of a computer, is configured to cause the computer to execute the modeling method for automotive component fault diagnosis presented in the embodiments of the present application.

Claims

1. An intelligent automobile component failure diagnosis modeling method, characterized by, The method includes: Step 1: Obtain vehicle driving history big data and maintenance record data; Step 2: Perform data aggregation and cleaning on historical data of the entire life cycle of automotive parts; the data aggregation is performed on a minute-by-minute basis, sorting the signals of each data field of the historical data in chronological order, grouping them on a minute-by-minute basis, and taking the average value of each minute group as the value of the current minute; Step 3: Segment and label the data using a single start-stop cycle during vehicle operation as a time sliding window; including... First, the aggregated data is sorted chronologically, and then grouped by single start / stop events. Secondly, the time series length of a single start-stop is fixed to a constant value L. When the time series length of a single start-stop exceeds L, only the data from the start time to time L is extracted as the time series data for this start-stop. When the time series length of a single start-stop does not exceed L, the time series data from the start time to the stop time remains unchanged, while the time series data from the stop time to time L is filled with 0 values. Finally, the data from each start-stop operation is labeled accordingly. Specifically, three time windows are set within the historical data of the entire life cycle of the components, namely T1, T2, and T3. The data labels in time window T1 are set as Label, and the data labels in time window T3 are set as the corresponding Label' according to the specific fault type. The data in time window T2 is discarded. In time window T1 and time window T3, the data is processed and labeled using a single start and stop of the vehicle as a sliding time window. Step 4: Construct a fault diagnosis model based on the LSTM time series model and the neural network model.

2. The intelligent automotive component failure diagnostic modeling method of claim 1, wherein, Step 1 specifically includes: Based on the maintenance record cards of each vehicle component, the fault type and corresponding repair time of each component are determined. Based on the fault type and repair time, historical data of the entire life cycle of the components during the vehicle's operation from the vehicle's manufacturing time to the repair time are extracted from the cloud platform. The data fields include the component itself, component load, vehicle driving external environment, and vehicle driving behavior CAN signal data.

3. The intelligent automotive component failure diagnostic modeling method of claim 1, wherein, The data cleaning includes removing outliers and filling in missing values ​​for each data field signal in the historical data of the entire life cycle of the component.

4. The intelligent automotive component failure diagnostic modeling method according to any one of claims 1-3, wherein, In step 4, a fault diagnosis model is constructed based on the LSTM time series model and the neural network model, and modeling analysis is performed. The time series segmented data generated in step 3 is sent into the constructed time series model for training and iteration.

5. An electronic device, comprising: At least one processor; The electronic device also includes a memory communicatively connected to at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform the intelligent automotive component fault diagnosis modeling method according to any one of claims 1 to 4.

6. A computer program product comprising a computer program, wherein, When executed by the computer's processor, the computer program is used to cause the computer to perform the intelligent automotive component fault diagnosis modeling method according to any one of claims 1 to 4.

Citation Information

Patent Citations

  • A real-time diagnosis system and diagnosis method for vehicle condition

    CN103576674B

  • Configurable fault diagnosis method and fault diagnosis system based on real-time vehicle condition

    CN114279718A

  • Self-adaptive aggregation processing method and system for automobile fault data

    CN114935457A